You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I believe that transfers done on the loopback interface are running slower than is possible. I'm only getting about 120-130 KB/sec, which amounts to roughly 1 Mbit/s (right?) or 1/100th of my "theoretical" throughput on my interface... and probably quite a bit lower than that for the loopback's theoretical limit, since any loopback speed should be higher than the actual physical interfaces are rated for, shouldn't it?
(More precisely the connection is to my local machine via it's Bonjour name "Firefly.local", which I believe merely runs through the loopback instead of being broadcast and re-recieved on the wire, right?)
I am only doing this while testing out Cyberduck, so it probably has limited impact in the real world... but nonetheless, it would seem this could be quite a bit faster than it is. I mean, even if such "*.local" connections do in fact hop out of and back into the physical interface, it should still be better than 1 Mbit/s on a 100 Mbit/s interface. Might be something to optimize in the networking code somewhere... when time allows, that is.
The text was updated successfully, but these errors were encountered:
Sorry, this may be related to #12134. It sounds like a very similar thing anyway.
I got ticket-happy today, and didn't do a search on this issue first. My apologies.
However, I'm not entirely sure that a 1% loopback speed is the same problem as a 10% local network speed. So, I'm leaving it open for your review. Thank you.
I believe that transfers done on the loopback interface are running slower than is possible. I'm only getting about 120-130 KB/sec, which amounts to roughly 1 Mbit/s (right?) or 1/100th of my "theoretical" throughput on my interface... and probably quite a bit lower than that for the loopback's theoretical limit, since any loopback speed should be higher than the actual physical interfaces are rated for, shouldn't it?
(More precisely the connection is to my local machine via it's Bonjour name "Firefly.local", which I believe merely runs through the loopback instead of being broadcast and re-recieved on the wire, right?)
I am only doing this while testing out Cyberduck, so it probably has limited impact in the real world... but nonetheless, it would seem this could be quite a bit faster than it is. I mean, even if such "*.local" connections do in fact hop out of and back into the physical interface, it should still be better than 1 Mbit/s on a 100 Mbit/s interface. Might be something to optimize in the networking code somewhere... when time allows, that is.
The text was updated successfully, but these errors were encountered: